composer/bin/fetch-spdx-identifiers

#!/usr/bin/env php
<?php

$identifiers = new SPDXLicenseIdentifiersOnline;
$printer = new JsonPrinter;
$printer->printStringArray($identifiers->getStrings());

/**
 * SPDX Identifier List from the registry.
 */
class SPDXLicenseIdentifiersOnline
{
    const REGISTRY = 'http://www.spdx.org/licenses/';
    const EXPRESSION = '//*[@typeof="spdx:License"]/code[@property="spdx:licenseId"]/text()';

    private $identifiers;

    /**
     * @return array
     */
    public function getStrings()
    {
        if ($this->identifiers) {
            return $this->identifiers;
        }
        $this->identifiers = $this->importNodesFromURL(
            self::REGISTRY,
            self::EXPRESSION
        );

        return $this->identifiers;
    }

    private function importNodesFromURL($url, $expressionTextNodes)
    {
        $doc = new DOMDocument();
        $doc->loadHTMLFile($url);
        $xp = new DOMXPath($doc);
        $codes = $xp->query($expressionTextNodes);
        if (!$codes) {
            throw new \Exception(sprintf('XPath query failed: %s', $expressionTextNodes));
        }
        if ($codes->length < 20) {
            throw new \Exception('Obtaining the license table failed, there can not be less than 20 identifiers.');
        }
        $identifiers = array();
        foreach ($codes as $code) {
            $identifiers[] = $code->nodeValue;
        }

        return $identifiers;
    }
}

/**
 * Print an array the way this script needs it.
 */
class JsonPrinter
{
    /**
     *
     * @param array $array
     */
    public function printStringArray(array $array)
    {
        $lines = array('');
        $line = &$lines[0];
        $last = count($array) - 1;
        foreach ($array as $item => $code) {
            $code = sprintf('"%s"%s', trim($code), $item === $last ? '' : ', ');
            $length = strlen($line) + strlen($code) - 1;
            if ($length > 76) {
                $line = rtrim($line);
                unset($line);
                $lines[] = $code;
                $line = &$lines[count($lines) - 1];
            } else {
                $line .= $code;
            }
        }
        $json = sprintf("[%s]", implode("\n    ", $lines));
        $json = str_replace(array("[\"", "\"]"), array("[\n    \"", "\"\n]"), $json);
        echo $json;
    }
}